Built-in Taints and Node Conditions
What You'll Learn
You'll discover how Kubernetes automatically applies taints to nodes based on real-time health conditions, protecting your workloads from unreliable infrastructure.
Understanding Automatic Node Taints
While you manually apply taints for custom scheduling rules, Kubernetes also automatically taints nodes when it detects problems. Think of these like warning signs that appear on a road when conditions become dangerous—no human needed to put them there.
When Kubernetes observes specific node conditions, it immediately applies corresponding taints to prevent new pods from scheduling there (and potentially evict existing ones).
Common Built-in Taints
NotReady Condition
When a node stops responding or becomes unhealthy, Kubernetes applies:
node.kubernetes.io/not-ready:NoExecute
This prevents new pods from scheduling and evicts pods without matching tolerations.
MemoryPressure Condition
When available memory drops critically low:
node.kubernetes.io/memory-pressure:NoSchedule
New pods won't schedule here unless they tolerate memory-constrained environments.
DiskPressure Condition
When disk space or inodes are running out:
node.kubernetes.io/disk-pressure:NoSchedule
Protects against pods failing due to insufficient disk resources.
Other Conditions
Additional automatic taints include node.kubernetes.io/network-unavailable, node.kubernetes.io/unschedulable, and node.kubernetes.io/pid-pressure.
Why This Matters
These automatic taints work with the NoExecute taint effect and toleration seconds you learned earlier. When a node becomes NotReady, pods without tolerations get evicted after their grace period expires—enabling automatic failover to healthy nodes without manual intervention.
Key Takeaway: Kubernetes automatically taints nodes experiencing NotReady, MemoryPressure, DiskPressure, and other critical conditions, creating self-healing infrastructure that prevents scheduling on unhealthy nodes without human intervention.